Tips for working with GPT-6 Astra: explicit autonomy prompts beat polite requests
daniel_mac8 · x · 2026-09-04
@angelbrodin shares a thread of practical tips for working with GPT-6 Astra: the model is an effective collaborator that proactively asks clarifying questions when more context could change the output — helpful, but it can stall when you expect it to make reasonable assumptions and keep going. Recommended fix: before kicking off long-running tasks, state explicitly "Infer my intent and task scope, bias towards action, make reasonable assumptions, work autonomously, and only pause for clearly destructive or irreversible actions." Also warns against phrasing commands as polite coworker requests ("could you…") when you actually mean "do the work."
More from coding & agent
- Intent update coordinates swarms of agents with workspaces across devices — LukeW · 2026-09-04
- LukeW's Intent update: multi-agent coordination, workspaces, multi-device runs — LukeW · 2026-09-04
- VC advice for agentic startups: pitch how your agent recovers from failures — atShruti · 2026-09-04
- WebMCP demo: an AI agent DJs in the browser with hundreds of direct tool calls — No_Guide_8697 · 2026-09-04
- Newsletters, not killer use cases, are what get users to trust agents with real email access — JanJanJaJa · 2026-09-04
- Kafka, Kafka Connect and Schema Registry exposed as native MCP tools — jkriket · 2026-09-04